Brief communication: CMIP6 does not suggest any circulation change over Greenland in summer by 2100

The Greenland blocking index (GBI), an indicator of the synoptic-scale circulation over Greenland, has been anomalously positive during summers since the late 1990s. Such changes in atmospheric circulation have led to an increase in Greenland summer temperatures, a decrease in cloud cover and greater surface melt. The GBI is therefore a key indicator of melting and surface mass balance variability over the Greenland ice sheet. However, the fifth phase of the Coupled Model Intercomparison Project (CMIP5) models do not represent any increase in GBI as suggested by observations. Until 2100, no significant 5 long-term trend in the GBI, and therefore no circulation changes, are projected. In this study the new generation of CMIP6 Earth-system models is evaluated in order to analyze the evolution of the future GBI. All CMIP5 and CMIP6 projections reveal the same trend towards a decrease of the GBI until 2100 and no model reproduces the strong increase in GBI observed over the last few decades. Significant melting events related to a highly positive GBI, as observed this summer 2019, are still not considered by CMIP6 models and therefore the projected surface melt increase of the ice sheet is likely to be underestimated 10 if such circulation changes persist in the next decades.

Despite the importance of such circulation changes, they are not represented by any of the CMIP5 Earth-system models (ESM) (Fettweis et al., 2013b;Hanna et al., 2018a). None of the models suggest any increase in Greenland blocking events by 2100, while some models even suggest moderate decreases in blocking. This model-observation disparity means that the melt increase over the Greenland ice sheet due to global warming could be underestimated by a factor of two, suggesting a potential underestimation of the future SMB decrease (Delhasse et al., 2018).

5
The latest Climate Model Intercomparison Project (CMIP6) provides new projections simulated by an improved generation of ESMs. The most important enhancements compared to CMIP5 are a higher resolution, a more sophisticated physics notably due the improvement of the coupling between the different components of the Earth-system, and better constrained concentrations of aerosols and other near-term climate forcings (Eyring et al., 2016;O'Neill et al., 2016;Voldoire et al., 2019). While the CMIP6 models are forced by similar atmospheric greenhouse gas scenarios, they are more sensitive than CMIP5 as they show 10 a stronger warming at the end of the Twenty First century. This could partly result from stronger cloud feedbacks (Andrews et al., 2019;Gettelman et al., 2019;Voldoire et al., 2019). For the Greenland ice sheet, a new investigation suggests a doubling of surface melt and a lengthening of the melt season in CMIP6 models relative to CMIP5 (Hofer et al. 2019 submitted). Considering the impact of such blocking events and the availability of new CMIP6 ESM projections, the aim here is to assess: 1) the ability of these simulations to represent the recent increase in Greenland blocking, and 2) whether such circulation/blocking changes are predicted from now until 2100.

Data and methodology
The summer (JJA) Greenland Blocking Index (GBI) is used to assess the representation in CMIP6 and CMIP5 models of the 5 recent summer blocking events observed over Greenland. GBI is defined as the area-weighted mean geopotantial height of the 500-hPa level (Z500) over the Greenland area 60-80 • N, 20-80 • W (GR) and is referred to as GB1 hereinafter (Hanna et al., 2016). In order to avoid the influence of the global temperature increase ( Figure 1) and study only the dynamic (and not thermal) atmospheric changes that occur above Greenland, the GB2 index (Eq. 1, Hanna et al., 2018a) is used here. GB2 is calculated as the normalized difference between GBI and the area-weighted mean Z500 over the northern hemisphere region 10 60-80 • N (NH).
The free-atmosphere temperature related to the GBI region (TA1 hereafter, Eq. 2) is the representation in a model of one of the factors driving and influenced by Z500 variability. Similarly to GBI, we used here TA2 which is defined as the difference of monthly mean temperature at 850, 700 and 500 hPa between the GR and NH areas (Eq. 2). TA2 is chosen for the same reason than GB2 and is used in its normalized form.
T A1 = (T 850 + T 700 + T 500) GR 3 T A2 = T A1 − (T 850 + T 700 + T 500) N H 3 TA2 and GB2 are based on summer CMIP5 andCMIP6 simulations (1950 -2100) and are compared with the same parameters based on NCEP/NCARv1 reanalysis (Kalnay et al., 1996, NCEP hereinafter) for the following recent period: 1950 -2019.  (2018a). A weighted running mean is applied by using a midpoint-centered 21-year running mean, which shortens the beginning and the end of series by 10 years, as shown in our figures.

Results
Figure 1 plotting TA1 vs GBI with respect to the reference period, shows that the GBI variability in the ESMs (both CMIP5 and 15 CMIP6) is closely linked to atmospheric (mid-troposphere) temperature changes. However, the observed GBI increases more strongly despite surface/lower atmosphere warming, due to a shift towards more favourable atmospheric dynamic (jet stream) precursors causing Greenland blocking events. Figure 1 also shows that the main difference between CMIP5 and CMIP6 is the stronger CMIP6 warming rate. The NCEP reanalysis shows a significant increase of GB2 (larger than the interannual variability) for the recent period, which is not represented by any of the CMIP5 and CMIP6 models (Figure 2). In CMIP5, the highest GB2 value does not even reach 1 standard deviation (std). Until 2020 two CMIP6 models (MRI ESM2-0 and EC-Earth3) reach 1 std of GB2 and oscillate until the end of the century with a decreasing trend. Until 2040, two other CMIP6 models (EC-Earth3-Veg and NESM3) oscillate near 1 std. Nevertheless, at the end of the century, all the ESMs project a decrease in GB2 and indicate no permanent circulation 5 changes or blocking increase, unlike what has recently occurred.
Similarly to GB2, the recent observed increase in TA2 (based on reanalysis) is mainly not represented by the CMIP6 ESMs ( Figure 3). Only one CMIP6 (MRI-ESM2-0) model suggests a temporary increase in TA2 during ten years as large as that of the NCEP record. The largest GB2 oscillations mentioned above are mainly due to the temperature variability and not caused by persistent circulation changes. This also suggests that the recent circulation anomalies are not due to natural variability. The 10 same models show as intense TA2 oscillations as GB2 whereas NCEP GB2 increase is more intense than TA2 which is not significant (< 1 std).

Conclusions
Blocking events have often characterized summers in Greenland since the late 1990s, leading to several melting records with 25 the latest one in 2019. This reflects particular atmospheric circulation conditions as gauged by the GBI index. The GBI has strongly increased since the late 1990s but yet the previous generation of ESM models (CMIP5) do not show such an increase (Hanna et al., 2018a). In this study we have used GB2 to evaluate the ability of the new CMIP6 models to represent recent variability of the atmospheric circulation over Greenland in summer, as well as its future evolution to 2100. We conclude that no circulation changes are represented for recent or future climates (1990s to 2100) by the CMIP6 models, and that the free-30 atmosphere temperature variability fully drives the GB2 changes in the ESM-based projections. This suggests that, according to CMIP5 and CMIP6 models, recent circulation anomalies are not due to natural variability.